Probably a bug in Garage SQLite

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
7 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Probably a bug in Garage SQLite

Herby Vojčík
Hello!

This testing code:

| databaseFile login accessor sqlString |
        databaseFile := Smalltalk imageDirectory asFileReference / 'play.sqlite'.
        login := Login new
                        database: SQLite3Platform new;
                        host: '';
                        port: '';
                        username: '';
                        password: '';
                        databaseName: databaseFile fullPath asZnUrl asString;
                        yourself.
        accessor := DatabaseAccessor forLogin: login.
        accessor login.
        sqlString := 'SELECT * FROM AGENT'.
        (accessor basicExecuteSQLString: sqlString) contents inspect.

fails (Pharo 6.1, GarageGlorp #stable loaded a week ago as a project
dependency) with:

Error: Unable to resolve external type: sqlite3

FFICallout(Object)>>error:
FFICallout>>resolveType:
FFICallout>>typeName:pointerArity:
FFICallout>>argName:indirectIndex:type:ptrArity:
FFIFunctionParser>>parseArgument
FFIFunctionParser>>parseArguments
FFIFunctionParser>>parseNamedFunction:
FFICalloutMethodBuilder>>parseSignature:
FFICalloutMethodBuilder>>generate
FFICalloutMethodBuilder>>build:
FFICalloutAPI>>function:module:
GASqlite3FFI(Object)>>ffiCall:module:
GASqlite3FFI>>apiErrorMessage:
GASqlite3FFI>>signal:with:on:
GASqlite3FFI>>checkForOk:on:
GASqlite3FFI>>prepare:on:with:
GASqlite3Statement>>prepare
GASqlite3Driver>>prepare:
GASqlite3ResultSet>>prepareStatement:
GASqlite3ResultSet>>execute:withCollection:
GASqlite3ResultSet>>execute:with:on:
GASqlite3Driver>>execute:with:
GASqlite3Driver>>execute:
GAGlorpDriver>>basicExecuteSQLString:
PharoDatabaseAccessor>>basicExecuteSQLString:
UndefinedObject>>DoIt
OpalCompiler>>evaluate
[..snip..]

Problem seems to be that in FFICallout >> resolveType:, the line

binding := resolver ffiBindingOf: name asSymbol.

actually produces nil, resolver being GASqlite3FFI class. If looking at
the class side of GaSqlite3FFI, there is no ffiBindingOf: at all; there is

nbBindingOf: aTypeName
        ^ TypeMap at: aTypeName ifAbsent: [ super nbBindingOf: aTypeName ]

though. If I copy near-mindlessly and add:

ffiBindingOf: aTypeName
        ^ TypeMap at: aTypeName ifAbsent: [ super ffiBindingOf: aTypeName ]

then the code above fails correctly with: 'no such table: AGENT'.

Am I missing some nb<->ffi bridge? Or is there a bug in Garage SQLite
driver?

Herby

Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Probably a bug in Garage SQLite

Stephane Ducasse-3
Hi Herby

On which version of Pharo are you trying? Because lot of changes
happened on Pharo 60 FFI.

Stef

On Sun, Aug 13, 2017 at 6:11 PM, Herby Vojčík <[hidden email]> wrote:

> Hello!
>
> This testing code:
>
> | databaseFile login accessor sqlString |
>         databaseFile := Smalltalk imageDirectory asFileReference /
> 'play.sqlite'.
>         login := Login new
>                         database: SQLite3Platform new;
>                         host: '';
>                         port: '';
>                         username: '';
>                         password: '';
>                         databaseName: databaseFile fullPath asZnUrl
> asString;
>                         yourself.
>         accessor := DatabaseAccessor forLogin: login.
>         accessor login.
>         sqlString := 'SELECT * FROM AGENT'.
>         (accessor basicExecuteSQLString: sqlString) contents inspect.
>
> fails (Pharo 6.1, GarageGlorp #stable loaded a week ago as a project
> dependency) with:
>
> Error: Unable to resolve external type: sqlite3
>
> FFICallout(Object)>>error:
> FFICallout>>resolveType:
> FFICallout>>typeName:pointerArity:
> FFICallout>>argName:indirectIndex:type:ptrArity:
> FFIFunctionParser>>parseArgument
> FFIFunctionParser>>parseArguments
> FFIFunctionParser>>parseNamedFunction:
> FFICalloutMethodBuilder>>parseSignature:
> FFICalloutMethodBuilder>>generate
> FFICalloutMethodBuilder>>build:
> FFICalloutAPI>>function:module:
> GASqlite3FFI(Object)>>ffiCall:module:
> GASqlite3FFI>>apiErrorMessage:
> GASqlite3FFI>>signal:with:on:
> GASqlite3FFI>>checkForOk:on:
> GASqlite3FFI>>prepare:on:with:
> GASqlite3Statement>>prepare
> GASqlite3Driver>>prepare:
> GASqlite3ResultSet>>prepareStatement:
> GASqlite3ResultSet>>execute:withCollection:
> GASqlite3ResultSet>>execute:with:on:
> GASqlite3Driver>>execute:with:
> GASqlite3Driver>>execute:
> GAGlorpDriver>>basicExecuteSQLString:
> PharoDatabaseAccessor>>basicExecuteSQLString:
> UndefinedObject>>DoIt
> OpalCompiler>>evaluate
> [..snip..]
>
> Problem seems to be that in FFICallout >> resolveType:, the line
>
> binding := resolver ffiBindingOf: name asSymbol.
>
> actually produces nil, resolver being GASqlite3FFI class. If looking at the
> class side of GaSqlite3FFI, there is no ffiBindingOf: at all; there is
>
> nbBindingOf: aTypeName
>         ^ TypeMap at: aTypeName ifAbsent: [ super nbBindingOf: aTypeName ]
>
> though. If I copy near-mindlessly and add:
>
> ffiBindingOf: aTypeName
>         ^ TypeMap at: aTypeName ifAbsent: [ super ffiBindingOf: aTypeName ]
>
> then the code above fails correctly with: 'no such table: AGENT'.
>
> Am I missing some nb<->ffi bridge? Or is there a bug in Garage SQLite
> driver?
>
> Herby
>

Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Probably a bug in Garage SQLite

Herby Vojčík
Stephane Ducasse wrote:

> Hi Herby
>
> On which version of Pharo are you trying? Because lot of changes
> happened on Pharo 60 FFI.
>
> Stef
>
> On Sun, Aug 13, 2017 at 6:11 PM, Herby Vojčík<[hidden email]>  wrote:
>> Hello!
>>
>> This testing code:
>>
>> | databaseFile login accessor sqlString |
>>          databaseFile := Smalltalk imageDirectory asFileReference /
>> 'play.sqlite'.
>>          login := Login new
>>                          database: SQLite3Platform new;
>>                          host: '';
>>                          port: '';
>>                          username: '';
>>                          password: '';
>>                          databaseName: databaseFile fullPath asZnUrl
>> asString;
>>                          yourself.
>>          accessor := DatabaseAccessor forLogin: login.
>>          accessor login.
>>          sqlString := 'SELECT * FROM AGENT'.
>>          (accessor basicExecuteSQLString: sqlString) contents inspect.
>>
>> fails (Pharo 6.1, GarageGlorp #stable loaded a week ago as a project

As I wrote here, Pharo 6.1.

>> dependency) with:
>>
>> Error: Unable to resolve external type: sqlite3
>>
>> FFICallout(Object)>>error:
>> FFICallout>>resolveType:
>> FFICallout>>typeName:pointerArity:
>> FFICallout>>argName:indirectIndex:type:ptrArity:
>> FFIFunctionParser>>parseArgument
>> FFIFunctionParser>>parseArguments
>> FFIFunctionParser>>parseNamedFunction:
>> FFICalloutMethodBuilder>>parseSignature:
>> FFICalloutMethodBuilder>>generate
>> FFICalloutMethodBuilder>>build:
>> FFICalloutAPI>>function:module:
>> GASqlite3FFI(Object)>>ffiCall:module:
>> GASqlite3FFI>>apiErrorMessage:
>> GASqlite3FFI>>signal:with:on:
>> GASqlite3FFI>>checkForOk:on:
>> GASqlite3FFI>>prepare:on:with:
>> GASqlite3Statement>>prepare
>> GASqlite3Driver>>prepare:
>> GASqlite3ResultSet>>prepareStatement:
>> GASqlite3ResultSet>>execute:withCollection:
>> GASqlite3ResultSet>>execute:with:on:
>> GASqlite3Driver>>execute:with:
>> GASqlite3Driver>>execute:
>> GAGlorpDriver>>basicExecuteSQLString:
>> PharoDatabaseAccessor>>basicExecuteSQLString:
>> UndefinedObject>>DoIt
>> OpalCompiler>>evaluate
>> [..snip..]
>>
>> Problem seems to be that in FFICallout>>  resolveType:, the line
>>
>> binding := resolver ffiBindingOf: name asSymbol.
>>
>> actually produces nil, resolver being GASqlite3FFI class. If looking at the
>> class side of GaSqlite3FFI, there is no ffiBindingOf: at all; there is
>>
>> nbBindingOf: aTypeName
>>          ^ TypeMap at: aTypeName ifAbsent: [ super nbBindingOf: aTypeName ]
>>
>> though. If I copy near-mindlessly and add:
>>
>> ffiBindingOf: aTypeName
>>          ^ TypeMap at: aTypeName ifAbsent: [ super ffiBindingOf: aTypeName ]
>>
>> then the code above fails correctly with: 'no such table: AGENT'.
>>
>> Am I missing some nb<->ffi bridge? Or is there a bug in Garage SQLite
>> driver?
>>
>> Herby

Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Probably a bug in Garage SQLite

Stephane Ducasse-3
Do you have the same bug in 5.0?
Because I do not know if garage was really tested on 60.

Stef

On Sun, Aug 13, 2017 at 8:41 PM, Herby Vojčík <[hidden email]> wrote:

> Stephane Ducasse wrote:
>>
>> Hi Herby
>>
>> On which version of Pharo are you trying? Because lot of changes
>> happened on Pharo 60 FFI.
>>
>> Stef
>>
>> On Sun, Aug 13, 2017 at 6:11 PM, Herby Vojčík<[hidden email]>  wrote:
>>>
>>> Hello!
>>>
>>> This testing code:
>>>
>>> | databaseFile login accessor sqlString |
>>>          databaseFile := Smalltalk imageDirectory asFileReference /
>>> 'play.sqlite'.
>>>          login := Login new
>>>                          database: SQLite3Platform new;
>>>                          host: '';
>>>                          port: '';
>>>                          username: '';
>>>                          password: '';
>>>                          databaseName: databaseFile fullPath asZnUrl
>>> asString;
>>>                          yourself.
>>>          accessor := DatabaseAccessor forLogin: login.
>>>          accessor login.
>>>          sqlString := 'SELECT * FROM AGENT'.
>>>          (accessor basicExecuteSQLString: sqlString) contents inspect.
>>>
>>> fails (Pharo 6.1, GarageGlorp #stable loaded a week ago as a project
>
>
> As I wrote here, Pharo 6.1.
>
>
>>> dependency) with:
>>>
>>> Error: Unable to resolve external type: sqlite3
>>>
>>> FFICallout(Object)>>error:
>>> FFICallout>>resolveType:
>>> FFICallout>>typeName:pointerArity:
>>> FFICallout>>argName:indirectIndex:type:ptrArity:
>>> FFIFunctionParser>>parseArgument
>>> FFIFunctionParser>>parseArguments
>>> FFIFunctionParser>>parseNamedFunction:
>>> FFICalloutMethodBuilder>>parseSignature:
>>> FFICalloutMethodBuilder>>generate
>>> FFICalloutMethodBuilder>>build:
>>> FFICalloutAPI>>function:module:
>>> GASqlite3FFI(Object)>>ffiCall:module:
>>> GASqlite3FFI>>apiErrorMessage:
>>> GASqlite3FFI>>signal:with:on:
>>> GASqlite3FFI>>checkForOk:on:
>>> GASqlite3FFI>>prepare:on:with:
>>> GASqlite3Statement>>prepare
>>> GASqlite3Driver>>prepare:
>>> GASqlite3ResultSet>>prepareStatement:
>>> GASqlite3ResultSet>>execute:withCollection:
>>> GASqlite3ResultSet>>execute:with:on:
>>> GASqlite3Driver>>execute:with:
>>> GASqlite3Driver>>execute:
>>> GAGlorpDriver>>basicExecuteSQLString:
>>> PharoDatabaseAccessor>>basicExecuteSQLString:
>>> UndefinedObject>>DoIt
>>> OpalCompiler>>evaluate
>>> [..snip..]
>>>
>>> Problem seems to be that in FFICallout>>  resolveType:, the line
>>>
>>> binding := resolver ffiBindingOf: name asSymbol.
>>>
>>> actually produces nil, resolver being GASqlite3FFI class. If looking at
>>> the
>>> class side of GaSqlite3FFI, there is no ffiBindingOf: at all; there is
>>>
>>> nbBindingOf: aTypeName
>>>          ^ TypeMap at: aTypeName ifAbsent: [ super nbBindingOf: aTypeName
>>> ]
>>>
>>> though. If I copy near-mindlessly and add:
>>>
>>> ffiBindingOf: aTypeName
>>>          ^ TypeMap at: aTypeName ifAbsent: [ super ffiBindingOf:
>>> aTypeName ]
>>>
>>> then the code above fails correctly with: 'no such table: AGENT'.
>>>
>>> Am I missing some nb<->ffi bridge? Or is there a bug in Garage SQLite
>>> driver?
>>>
>>> Herby
>
>

Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Probably a bug in Garage SQLite

Herby Vojčík
Stephane Ducasse wrote:
> Do you have the same bug in 5.0?
> Because I do not know if garage was really tested on 60.

Don't know (don't have it), but last commit in Garage-SQLite
GuillermoPolito.26 from 2016. The mentioned #nbBindingOf: is sent at no
place at all in the image (except supersend in the lone method itself),
so it definitely looks like it need to rename to ffiBindingOf: for 60.

Herby

>
> Stef
>
> On Sun, Aug 13, 2017 at 8:41 PM, Herby Vojčík<[hidden email]>  wrote:
>> Stephane Ducasse wrote:
>>> Hi Herby
>>>
>>> On which version of Pharo are you trying? Because lot of changes
>>> happened on Pharo 60 FFI.
>>>
>>> Stef
>>>
>>> On Sun, Aug 13, 2017 at 6:11 PM, Herby Vojčík<[hidden email]>   wrote:
>>>> Hello!
>>>>
>>>> This testing code:
>>>>
>>>> | databaseFile login accessor sqlString |
>>>>           databaseFile := Smalltalk imageDirectory asFileReference /
>>>> 'play.sqlite'.
>>>>           login := Login new
>>>>                           database: SQLite3Platform new;
>>>>                           host: '';
>>>>                           port: '';
>>>>                           username: '';
>>>>                           password: '';
>>>>                           databaseName: databaseFile fullPath asZnUrl
>>>> asString;
>>>>                           yourself.
>>>>           accessor := DatabaseAccessor forLogin: login.
>>>>           accessor login.
>>>>           sqlString := 'SELECT * FROM AGENT'.
>>>>           (accessor basicExecuteSQLString: sqlString) contents inspect.
>>>>
>>>> fails (Pharo 6.1, GarageGlorp #stable loaded a week ago as a project
>>
>> As I wrote here, Pharo 6.1.
>>
>>
>>>> dependency) with:
>>>>
>>>> Error: Unable to resolve external type: sqlite3
>>>>
>>>> FFICallout(Object)>>error:
>>>> FFICallout>>resolveType:
>>>> FFICallout>>typeName:pointerArity:
>>>> FFICallout>>argName:indirectIndex:type:ptrArity:
>>>> FFIFunctionParser>>parseArgument
>>>> FFIFunctionParser>>parseArguments
>>>> FFIFunctionParser>>parseNamedFunction:
>>>> FFICalloutMethodBuilder>>parseSignature:
>>>> FFICalloutMethodBuilder>>generate
>>>> FFICalloutMethodBuilder>>build:
>>>> FFICalloutAPI>>function:module:
>>>> GASqlite3FFI(Object)>>ffiCall:module:
>>>> GASqlite3FFI>>apiErrorMessage:
>>>> GASqlite3FFI>>signal:with:on:
>>>> GASqlite3FFI>>checkForOk:on:
>>>> GASqlite3FFI>>prepare:on:with:
>>>> GASqlite3Statement>>prepare
>>>> GASqlite3Driver>>prepare:
>>>> GASqlite3ResultSet>>prepareStatement:
>>>> GASqlite3ResultSet>>execute:withCollection:
>>>> GASqlite3ResultSet>>execute:with:on:
>>>> GASqlite3Driver>>execute:with:
>>>> GASqlite3Driver>>execute:
>>>> GAGlorpDriver>>basicExecuteSQLString:
>>>> PharoDatabaseAccessor>>basicExecuteSQLString:
>>>> UndefinedObject>>DoIt
>>>> OpalCompiler>>evaluate
>>>> [..snip..]
>>>>
>>>> Problem seems to be that in FFICallout>>   resolveType:, the line
>>>>
>>>> binding := resolver ffiBindingOf: name asSymbol.
>>>>
>>>> actually produces nil, resolver being GASqlite3FFI class. If looking at
>>>> the
>>>> class side of GaSqlite3FFI, there is no ffiBindingOf: at all; there is
>>>>
>>>> nbBindingOf: aTypeName
>>>>           ^ TypeMap at: aTypeName ifAbsent: [ super nbBindingOf: aTypeName
>>>> ]
>>>>
>>>> though. If I copy near-mindlessly and add:
>>>>
>>>> ffiBindingOf: aTypeName
>>>>           ^ TypeMap at: aTypeName ifAbsent: [ super ffiBindingOf:
>>>> aTypeName ]
>>>>
>>>> then the code above fails correctly with: 'no such table: AGENT'.
>>>>
>>>> Am I missing some nb<->ffi bridge? Or is there a bug in Garage SQLite
>>>> driver?
>>>>
>>>> Herby
>>
>


Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Probably a bug in Garage SQLite

Stephane Ducasse-3
ah yes.
probably a missing rename during the migration to the new ffi.

Stef

On Sun, Aug 13, 2017 at 9:04 PM, Herby Vojčík <[hidden email]> wrote:

> Stephane Ducasse wrote:
>>
>> Do you have the same bug in 5.0?
>> Because I do not know if garage was really tested on 60.
>
>
> Don't know (don't have it), but last commit in Garage-SQLite
> GuillermoPolito.26 from 2016. The mentioned #nbBindingOf: is sent at no
> place at all in the image (except supersend in the lone method itself), so
> it definitely looks like it need to rename to ffiBindingOf: for 60.
>
> Herby
>
>
>>
>> Stef
>>
>> On Sun, Aug 13, 2017 at 8:41 PM, Herby Vojčík<[hidden email]>  wrote:
>>>
>>> Stephane Ducasse wrote:
>>>>
>>>> Hi Herby
>>>>
>>>> On which version of Pharo are you trying? Because lot of changes
>>>> happened on Pharo 60 FFI.
>>>>
>>>> Stef
>>>>
>>>> On Sun, Aug 13, 2017 at 6:11 PM, Herby Vojčík<[hidden email]>   wrote:
>>>>>
>>>>> Hello!
>>>>>
>>>>> This testing code:
>>>>>
>>>>> | databaseFile login accessor sqlString |
>>>>>           databaseFile := Smalltalk imageDirectory asFileReference /
>>>>> 'play.sqlite'.
>>>>>           login := Login new
>>>>>                           database: SQLite3Platform new;
>>>>>                           host: '';
>>>>>                           port: '';
>>>>>                           username: '';
>>>>>                           password: '';
>>>>>                           databaseName: databaseFile fullPath asZnUrl
>>>>> asString;
>>>>>                           yourself.
>>>>>           accessor := DatabaseAccessor forLogin: login.
>>>>>           accessor login.
>>>>>           sqlString := 'SELECT * FROM AGENT'.
>>>>>           (accessor basicExecuteSQLString: sqlString) contents inspect.
>>>>>
>>>>> fails (Pharo 6.1, GarageGlorp #stable loaded a week ago as a project
>>>
>>>
>>> As I wrote here, Pharo 6.1.
>>>
>>>
>>>>> dependency) with:
>>>>>
>>>>> Error: Unable to resolve external type: sqlite3
>>>>>
>>>>> FFICallout(Object)>>error:
>>>>> FFICallout>>resolveType:
>>>>> FFICallout>>typeName:pointerArity:
>>>>> FFICallout>>argName:indirectIndex:type:ptrArity:
>>>>> FFIFunctionParser>>parseArgument
>>>>> FFIFunctionParser>>parseArguments
>>>>> FFIFunctionParser>>parseNamedFunction:
>>>>> FFICalloutMethodBuilder>>parseSignature:
>>>>> FFICalloutMethodBuilder>>generate
>>>>> FFICalloutMethodBuilder>>build:
>>>>> FFICalloutAPI>>function:module:
>>>>> GASqlite3FFI(Object)>>ffiCall:module:
>>>>> GASqlite3FFI>>apiErrorMessage:
>>>>> GASqlite3FFI>>signal:with:on:
>>>>> GASqlite3FFI>>checkForOk:on:
>>>>> GASqlite3FFI>>prepare:on:with:
>>>>> GASqlite3Statement>>prepare
>>>>> GASqlite3Driver>>prepare:
>>>>> GASqlite3ResultSet>>prepareStatement:
>>>>> GASqlite3ResultSet>>execute:withCollection:
>>>>> GASqlite3ResultSet>>execute:with:on:
>>>>> GASqlite3Driver>>execute:with:
>>>>> GASqlite3Driver>>execute:
>>>>> GAGlorpDriver>>basicExecuteSQLString:
>>>>> PharoDatabaseAccessor>>basicExecuteSQLString:
>>>>> UndefinedObject>>DoIt
>>>>> OpalCompiler>>evaluate
>>>>> [..snip..]
>>>>>
>>>>> Problem seems to be that in FFICallout>>   resolveType:, the line
>>>>>
>>>>> binding := resolver ffiBindingOf: name asSymbol.
>>>>>
>>>>> actually produces nil, resolver being GASqlite3FFI class. If looking at
>>>>> the
>>>>> class side of GaSqlite3FFI, there is no ffiBindingOf: at all; there is
>>>>>
>>>>> nbBindingOf: aTypeName
>>>>>           ^ TypeMap at: aTypeName ifAbsent: [ super nbBindingOf:
>>>>> aTypeName
>>>>> ]
>>>>>
>>>>> though. If I copy near-mindlessly and add:
>>>>>
>>>>> ffiBindingOf: aTypeName
>>>>>           ^ TypeMap at: aTypeName ifAbsent: [ super ffiBindingOf:
>>>>> aTypeName ]
>>>>>
>>>>> then the code above fails correctly with: 'no such table: AGENT'.
>>>>>
>>>>> Am I missing some nb<->ffi bridge? Or is there a bug in Garage SQLite
>>>>> driver?
>>>>>
>>>>> Herby
>>>
>>>
>>
>
>

Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Probably a bug in Garage SQLite

Herby Vojčík
Stephane Ducasse wrote:
> ah yes.
> probably a missing rename during the migration to the new ffi.
>
> Stef

Unfortunately there's more:

PharoDatabaseAccessor(DatabaseAccessor)>>handleError:for:
[ :ex | self handleError: ex for: command ] in [ | result |
self checkPermissionFor: command.
result := [ (self useBinding and: [ command useBinding ])
        ifTrue: [ command executeBoundIn: self ]
        ifFalse: [ command executeUnboundIn: self ] ]
        on: Dialect error
        do: [ :ex | self handleError: ex for: command ].
aBoolean
        ifTrue: [ result ]
        ifFalse: [ result upToEnd ] ] in
PharoDatabaseAccessor(DatabaseAccessor)>>executeCommand:returnCursor:
BlockClosure>>cull:
Context>>evaluateSignal:
Context>>handleSignal:
MessageNotUnderstood(Exception)>>signal
ExternalData(Object)>>doesNotUnderstand: #pointerAt:
FFIExternalArray class>>fromPointer:type:size:
GASqlite3FFI>>blobFrom:at:
GASqlite3Statement>>byteArrayAt:
GASqlite3Statement>>valueOfColumn:
GASqlite3ResultSet>>next
GASqlite3ResultSet>>rows
GAGlorpDriver>>basicExecuteSQLString:
PharoDatabaseAccessor>>basicExecuteSQLString:
PharoDatabaseAccessor>>executeCommandUnbound:
[..snip..]

Now ExternalData does not understand #pointerAt: (only ByteArray and
FFIExternalStructureReferenceHandle do, don't know how to fix this
quickly... :-( )

Herby

>
> On Sun, Aug 13, 2017 at 9:04 PM, Herby Vojčík<[hidden email]>  wrote:
>> Stephane Ducasse wrote:
>>> Do you have the same bug in 5.0?
>>> Because I do not know if garage was really tested on 60.
>>
>> Don't know (don't have it), but last commit in Garage-SQLite
>> GuillermoPolito.26 from 2016. The mentioned #nbBindingOf: is sent at no
>> place at all in the image (except supersend in the lone method itself), so
>> it definitely looks like it need to rename to ffiBindingOf: for 60.
>>
>> Herby
>>
>>
>>> Stef
>>>
>>> On Sun, Aug 13, 2017 at 8:41 PM, Herby Vojčík<[hidden email]>   wrote:
>>>> Stephane Ducasse wrote:
>>>>> Hi Herby
>>>>>
>>>>> On which version of Pharo are you trying? Because lot of changes
>>>>> happened on Pharo 60 FFI.
>>>>>
>>>>> Stef
>>>>>
>>>>> On Sun, Aug 13, 2017 at 6:11 PM, Herby Vojčík<[hidden email]>    wrote:
>>>>>> Hello!
>>>>>>
>>>>>> This testing code:
>>>>>>
>>>>>> | databaseFile login accessor sqlString |
>>>>>>            databaseFile := Smalltalk imageDirectory asFileReference /
>>>>>> 'play.sqlite'.
>>>>>>            login := Login new
>>>>>>                            database: SQLite3Platform new;
>>>>>>                            host: '';
>>>>>>                            port: '';
>>>>>>                            username: '';
>>>>>>                            password: '';
>>>>>>                            databaseName: databaseFile fullPath asZnUrl
>>>>>> asString;
>>>>>>                            yourself.
>>>>>>            accessor := DatabaseAccessor forLogin: login.
>>>>>>            accessor login.
>>>>>>            sqlString := 'SELECT * FROM AGENT'.
>>>>>>            (accessor basicExecuteSQLString: sqlString) contents inspect.
>>>>>>
>>>>>> fails (Pharo 6.1, GarageGlorp #stable loaded a week ago as a project
>>>>
>>>> As I wrote here, Pharo 6.1.
>>>>
>>>>
>>>>>> dependency) with:
>>>>>>
>>>>>> Error: Unable to resolve external type: sqlite3
>>>>>>
>>>>>> FFICallout(Object)>>error:
>>>>>> FFICallout>>resolveType:
>>>>>> FFICallout>>typeName:pointerArity:
>>>>>> FFICallout>>argName:indirectIndex:type:ptrArity:
>>>>>> FFIFunctionParser>>parseArgument
>>>>>> FFIFunctionParser>>parseArguments
>>>>>> FFIFunctionParser>>parseNamedFunction:
>>>>>> FFICalloutMethodBuilder>>parseSignature:
>>>>>> FFICalloutMethodBuilder>>generate
>>>>>> FFICalloutMethodBuilder>>build:
>>>>>> FFICalloutAPI>>function:module:
>>>>>> GASqlite3FFI(Object)>>ffiCall:module:
>>>>>> GASqlite3FFI>>apiErrorMessage:
>>>>>> GASqlite3FFI>>signal:with:on:
>>>>>> GASqlite3FFI>>checkForOk:on:
>>>>>> GASqlite3FFI>>prepare:on:with:
>>>>>> GASqlite3Statement>>prepare
>>>>>> GASqlite3Driver>>prepare:
>>>>>> GASqlite3ResultSet>>prepareStatement:
>>>>>> GASqlite3ResultSet>>execute:withCollection:
>>>>>> GASqlite3ResultSet>>execute:with:on:
>>>>>> GASqlite3Driver>>execute:with:
>>>>>> GASqlite3Driver>>execute:
>>>>>> GAGlorpDriver>>basicExecuteSQLString:
>>>>>> PharoDatabaseAccessor>>basicExecuteSQLString:
>>>>>> UndefinedObject>>DoIt
>>>>>> OpalCompiler>>evaluate
>>>>>> [..snip..]
>>>>>>
>>>>>> Problem seems to be that in FFICallout>>    resolveType:, the line
>>>>>>
>>>>>> binding := resolver ffiBindingOf: name asSymbol.
>>>>>>
>>>>>> actually produces nil, resolver being GASqlite3FFI class. If looking at
>>>>>> the
>>>>>> class side of GaSqlite3FFI, there is no ffiBindingOf: at all; there is
>>>>>>
>>>>>> nbBindingOf: aTypeName
>>>>>>            ^ TypeMap at: aTypeName ifAbsent: [ super nbBindingOf:
>>>>>> aTypeName
>>>>>> ]
>>>>>>
>>>>>> though. If I copy near-mindlessly and add:
>>>>>>
>>>>>> ffiBindingOf: aTypeName
>>>>>>            ^ TypeMap at: aTypeName ifAbsent: [ super ffiBindingOf:
>>>>>> aTypeName ]
>>>>>>
>>>>>> then the code above fails correctly with: 'no such table: AGENT'.
>>>>>>
>>>>>> Am I missing some nb<->ffi bridge? Or is there a bug in Garage SQLite
>>>>>> driver?
>>>>>>
>>>>>> Herby
>>>>
>>
>


Loading...